Hi

I have not done a time series analysis before. We have 50 patients that have been vaccinated and we are observing 1 baseline timepoint and 3 timepoints post original vaccination in the microarray experiment. We just want to find differentially expressed genes across these timepoints. So far I have run a paired t-test in limma which is fine, but myself and my colleges want to try more techniques. It seems people use regression type approaches to answer this as in maSigPro or there is the timecourse package I have found which seems to use some kind of special ANOVA style approach. However, I am wondering about how applicable these methods are to such a small number of timepoints such as what we have? Are these the preferred methods? I am also wondering about the best clustering approach for any differential expression during a time course? Any words of wisdom would be appreciated.

Thanks,

Chris

It's easy to run a time-course regression in limma -- no need to restrict to pairwise t-tests -- but there usually isn't that much to be gained with only 4 timepoints.